Thursday, May 13, 2021

Car fleet proportions, Part 2

 This post is a follow-on to the recent post about proportions of a freight car fleet, a much more general topic (you can review that post at this link: ). In the present post, I want to turn to the issue, not of fleet size, but of what cars are in each fleet. The previous post’s graphs were for entire car fleets (minus coal, ore and ballast cars). Now let’s delver further into details. 

I would begin with the national car fleet, as described in ICC statistics for Dec. 30, 1950. The table shown below was published in the 1953 Car Builders’ Cyclopedia, page 69.

What I suppose we could call the “bottom line” is at lower right, the total U.S. freight car fleet size of a little over 2 million cars. These statistics unfortunately combine gondola and hopper cars, both important categories. I only have data for Class I railroads (one of the columns above), but for 1950, there were 556,000 hoppers, 23,000 covered hoppers, and 285,000 gondolas. We can use this ratio, 556 / 865, to estimate that in the above table, the 882,000 combined hoppers and gondolas comprised 572,000 hoppers.

But other categories are clearly called out in the table, and we can recognize that the largest category is box cars (including auto cars), at about 720,000, followed by hoppers and gondolas. With that bottom-line number, we can make a bar graph of the national car fleet at the end of 1950.

Keep in mind, looking at this graph, that no railroad or any car owner could match this graph exactly. As the table above shows, nearly all tank cars and refrigerator cars were in private ownership. Moreover, each railroad owned a fleet of freight cars suited to its territory and its traffic. The graph of the car fleet for one’s layout might look like this, but no individual car owner could do so.

This is an interesting piece of information, the composition of the national fleet, but how do we do the same for individual railroads? We turn to the Official Railway Equipment Register, or ORER, in this case for April 1950 (the issue I have). We can quickly assemble bar graphs like the one shown above, but for individual railroads. 

I will begin with the Pennsylvania. We saw in the previous post that their car fleet was the largest in the U.S., with or without hopper cars. Here is a graph of the constituents of the PRR fleet.  I’ve retained the order of car types from the graph above.

Let us look for a second at just the Pennsylvania box cars. In 1950, they were more than 8 percent of the entire national fleet of box cars, so on many layouts, you might expect 8 percent of the foreign road box cars to be PRR cars. (Note that I separate foreign cars from home-road cars; the issue of home-road car percentages is an interesting but separate one, as I’ve described: .) 

 What that 8 percent would mean can be shown with a numerical example. Let’s imagine that you have, or plan to have, a freight car fleet for your layout that will include 150 box cars that can be foreign-road cars (this is a big number, but just an example). The data above then suggest that 12 of your 150 foreign-road box cars will be PRR cars.

What about a sizeable road, though the smallest one in the bar graph of railroad fleets in the previous post, the Rock Island (formally, Chicago, Rock Island & Pacific)? Let’s look at that. First, here’s a graph like the one above, dramatizing how the Rock Island fleet was dominated by box cars. (Many other Western railroad fleets were also dominated by box cars.) Unless you model a road adjoining the Rock Island, you are only likely to see box cars as Rock Island foreign cars. Note also how different was the composition of the Rock Island fleet, compared to the PRR graph shown above.

In 1950, the presumably free-running Rock Island box cars, AAR type XM, numbered a bit over 17,000 cars. The national fleet of such cars was about 720,000 cars, making the Rock Island fleet about 2.3 percent of all box cars. In model terms, that might mean that if my model railroad has 150 foreign-road box cars, about 3 should carry RI reporting marks. 

Now I will show just one more example, the Southern Railway, which is interesting because of the substantial number of flat cars, well above the national average. Moreover, the ORER shows us that fully two-thirds of these were AAR type LP flat cars, meaning equipped for pulpwood service. This was a regional emphasis (the Seaboard fleet was similar in this regard), but the difference emphasizes the need to understand each railroad separately. And one might well decide that those pulpwood flats would not travel far beyond the rails of the Southern or its neighbors.

I will continue with some more observations on individual railroad car fleets in a future post.

Tony Thompson


  1. Tony,
    I remember getting into this topic after reading Bruce Chubb's book that discussed it. Per your comment, "if my model railroad has 150 foreign-road box cars, about 3 should carry RI reporting marks," I assume that you will discuss the impact of the specific industries on your RR and the proximity of other major roads to yours. Given your location and being completely surrounded by several major western roads, I would not be surprised if a RI XM only showed up once in awhile. Looking forward to Parts 3 and beyond.
    Lou Adler

    1. Lou, you are right that proximity of railroads plays some role, but the entire point of the Gilbert-Nelson idea is that free-running cars really run freely everywhere. Those 3 RI cars should show up on a layout anywhere (for modeled locales that don't interchange with the RI or actually ARE the RI).
      Tony Thompson

    2. Hi Tony:
      Good blog and some good thoughts. However, the Gilbert-Nelson theory perhaps does not account for the existence of the car service rules, which will tend to skew the "free running" cars somewhat toward their home districts due to both the dispatch and empty requirements. While I agree that the car service rules were often honored in their breach, particularly during periods of car shortage, in the 1950s an AAR study showed they were followed about 50-60% of the time, which should "starve" districts further away from the home district(s) compared to a completely "free-ranging" model. This effect may or may not be great enough to be noticeable in a typical model railroader car fleet.
      Thanks for the great thought piece on this topic and keep on writing!
      Eric Hiser

    3. Eric, thanks for the thoughtful comments. I agree that perhaps Gilbert-Nelson SHOULD account for Car Service Rules. But the point to be grasped is that the available data DO support an unmodified Gilber-Nelson hypothesis. We don't need to modify it.

      I sure wish we had hundreds of conductor time books to serve as data to test Gilbert-Nelson in much more detail, for railroads large and small and for all regions of the country. Maybe if we had all those data, we would find needs to modify the original G-N idea. But so far, not the case.
      Tony Thompson

  2. I remember over 20 years ago never wanting to miss a PCR Convention whenever you gave this clinic. It was way ahead of its time then and it is still great info that remains ahead of its time. Thank you for everything you have done to promote model RRing and the Southern Pacific.